Support "mean" in describe for bools

[Julian-J-S]

Description

Data often contains columns with flags (yes/no, true/false).

Getting a quick overview of the column requires actually only 2 things

1. amount of valid/invalid values -> available! (count/null_count) ✅
2. ratio between yes/no values -> NOT available 🟥

This can be easily solved by adding the "mean" value of the boolean column

• [True, False, False, None] -> 0.33 (33% True)
• [True, True, True, None] -> 1.00 (100% True)
• [False, False, False, None] -> 0.0 (0% True)
• [None, None, None, None] -> 0.0 (0% True)

Current State:

Problem: almost useless for bool columns. Only information is the null_count

pl.DataFrame({"bool": [True, False, False, None]}).describe()

shape: (9, 2)
┌────────────┬───────┐
│ describe   ┆ bool  │
│ ---        ┆ ---   │
│ str        ┆ str   │
╞════════════╪═══════╡
│ count      ┆ 3     │
│ null_count ┆ 1     │
│ mean       ┆ null  │ <<<<< this would be really usefull to get the Yes/No-Ratio
│ std        ┆ null  │
│ min        ┆ False │
│ 25%        ┆ null  │
│ 50%        ┆ null  │
│ 75%        ┆ null  │
│ max        ┆ True  │
└────────────┴───────┘

[alexander-beedie]

I'm sold; it seems genuinely useful as an indicator.

[taki-mekhalfa]

Happy to take this one